General Motors Is Cutting Its Development Cycles in Half

General Motors is deploying artificial intelligence and simulation to cut vehicle development cycles in half, a move aimed at matching the speed of Chinese automakers such as BYD that can bring new models to market in two years or less [1]. The effort is led by Sterling Anderson, GM’s chief product officer, who previously led development teams for Tesla’s Autopilot and the Model X before cofounding the autonomous trucking company Aurora Innovation [1][7]. Anderson joined GM last June with a compensation package valued at $40 million [1]. He describes the company’s current approach as a third epoch of design, following millennia of empirical methods and the later introduction of computer-aided tools [1]. GM has already applied the new process to the electric GMC Hummer, which moved from initial designs to showrooms in two years, compared with a typical four- to five-year product cycle [1]. The company’s goal is to extend that pace across a full range of vehicle and technology programs. “We’re not there yet, but give us a minute,” Anderson said [1]. The automaker, headquartered in Detroit, Michigan, has been the largest in the United States by total sales and was the world’s largest for 77 years before losing the title to Toyota in 2008 [2]. Founded in 1908 as a holding company for Buick, GM now owns four core brands — Chevrolet, Buick, GMC, and Cadillac — and operates manufacturing plants in eight countries [2][3]. Jason Fischer, GM’s executive director of virtual integration engineering, said the company’s proprietary simulation environment allows engineers to simultaneously develop and optimize hardware and software before physical prototypes are built [1]. In one demonstration, a digitally rendered Cadillac Escalade IQ was run through a double-lane-change avoidance maneuver. Fischer said the system enables “full, virtual calibrations prior to a vehicle ever being built” [1]. Front-end crash simulations that once required 15 hours of computing can now be completed in less than one minute using an AI method based on probabilities, Fischer said [1]. Anderson added that the company is approaching roughly 2 million simulation runs per week. “We can simulate 100 days of driving in a day,” he said, “which helps us probe edge cases that would be dangerous or impractical to reproduce physically” [1]. GM is also using the tools for work beyond passenger vehicles, including a next-generation lunar rover for NASA’s Artemis mission, which aims to land astronauts on the moon in 2028 [1]. Fischer noted that engineers can alter gravity by adjusting physics in the software. “Our engineers in a room in Michigan can simulate real-world driving conditions to develop tires for the lunar environment,” he said [1]. The push to accelerate development cycles mirrors a broader industrial shift toward programmable automation that began with the first numerically controlled machine tools in the 1940s and 1950s [4]. GM’s application of AI to structural design has already produced components such as a strut bracket for the Chevrolet Corvette’s hatch lid that is lighter, stiffer, and more durable than the original part [1].